Re: *printf_r and reentrancy

[Jeff Johnston <jjohnstn at redhat dot com>]

Eric Blake wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> I noticed that various *printf_r routines all get bundled through
> _vfprintf_r, which in turn calls __swsetup (via the macro cantwrite),
> which in turn can call __smakebuf and _malloc_r(_REENT).  But isn't this
> design broken, if an alternate reentrancy object (rather than _REENT) was
> originally passed to the *printf_r routine?  In other words, should
> cantwrite, __swsetup, and __smakebuf all be converted to pass a reentrancy
> argument around?

Yes, you are right for non-dynamic-reent configurations.  An error on 
the allocation will cause the errno to be set for _REENT which is 
incorrect except in the case of dynamic reent.  This needs to be fixed. 
 It's a bit of a can of worms.  Do you want to make the changes or 
would you prefer I do it?

-- Jeff J.